Andrey, please take look at comment below. On 2018-02-15 04:27, Andrei Vagin wrote: > [This sender failed our fraud detection checks and may not be who they appear > to be. Learn about spoofing at http://aka.ms/LearnAboutSpoofing] > > preq->io_count is incrimented before calling aio_kernel_submit() > and it is decrimented in kaio_rw_aio_complete(). > > But if aio_kernel_submit() failed , preq->io_count has to be decrimented > before exiting from the function. > --- > drivers/block/ploop/io_kaio.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/drivers/block/ploop/io_kaio.c b/drivers/block/ploop/io_kaio.c > index 2e48d13..6d922f5 100644 > --- a/drivers/block/ploop/io_kaio.c > +++ b/drivers/block/ploop/io_kaio.c > @@ -686,6 +686,7 @@ kaio_io_page(struct ploop_io * io, int op, struct > ploop_request * preq, > err, (op == IOCB_CMD_WRITE_ITER) ? "WRITE" : "READ", > preq->eng_state, preq->state, pos); > PLOOP_REQ_SET_ERROR(preq, err); > + ploop_complete_io_request(preq);
Kostja pointed me this patch for vz6. vz6 seems is affected too, however I think it's better to use atomic_dec(&preq->io_count) here: it should balance atomic_inc called before aio_kernel_submit() and should not double ploop_complete_io_request called below. How do you think? > } > > out: > -- > 1.8.3.1 > > > _______________________________________________ > Devel mailing list > Devel@openvz.org > https://lists.openvz.org/mailman/listinfo/devel > _______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel
